Research map / Image generation / Text-to-image diffusion
Constructing Dreams Using Generative AI
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Presents "Dreaming with AI", a 1h20m high-school workshop where 34 students used Stable Diffusion (Dream Studio) to visualize imagined future selves while learning about generative AI capabilities, limitations, and ethics.
Problem
K-12 students have little understanding of how generative AI tools work and their ethical implications, and few curricula exist for generative AI literacy.
Approach
Combines constructionist learning, identity expression (imagined future selves as text-to-image prompts), and critical debate to teach generative AI technical and ethical concepts to high schoolers.
Key results
- Students generated an average of 12.6 images each (min 4, max 31), and all finished with prompts longer than their initial prompts.
- Of 34 students, only 6 reported familiarity with generative AI, while 29 knew ChatGPT.
- Students could partially reach ethical objective E1, identifying algorithmic bias, fake media, plagiarism and job replacement, but did not identify data privacy.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use a future-identity expression activity with text-to-image generation and peer sharing | Constructing Dreams workshop · High school students, two workshops | creative learning objectives achieved | Helped |
| Use a future-identity expression activity with text-to-image generation and peer sharing | Constructing Dreams workshop · High school students, two workshops | prompt iteration: 12.6 images | Helped |
| Use a future-identity expression activity followed by reflection | Constructing Dreams workshop · High school students, two workshops | identifying patterns in training datasets | Mixed |
Surprising findings
- Students did not connect generated image patterns to training data patterns on their own (T4), only after instructor probing.
- Students in the US were much more positive about generative AI in classrooms, while students in India primarily viewed it as plagiarism.
Limitations
- No formal assessment of student learning; inferences about all students' learning cannot be made.
- Small, two-group sample from specific high-school programs (US and India), with incomplete response recording.
- Data privacy was not identified as a harm by students.
Experimental setup
Metrics: student participation counts qualitative thematic observations of discussions
Builds on: AI4K12 big ideas Identity Construction Environments constructionism
Closest papers by meaning
- A Picture Is Worth a Thousand Words: Co-designing Text-to-Image Generation Learning Materials for K-12 with Educators (AAAI 2024 · similarity 0.85)
- AI, Ethics, and Education: The Pioneering Path of Sidekick Academy (AAAI 2024 · similarity 0.84)
- From Consumers to Critical Users: Prompty, an AI Literacy Tool for High School Students (AAAI 2024 · similarity 0.84)
- Designing Characters with AI: An Art & AI Learning Activity (AAAI 2025 · similarity 0.84)
- Advancing Research on Equitable AI Education Through a Focus on Implementation: Insights from a Middle School Computer Vision Module Beta-Test (AAAI 2025 · similarity 0.83)
- Design and Evaluation of a Generative Artificial Intelligence-Based Tool for Students with Learning Disabilities in Kenya (AAAI 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.